Estimation of relative distance from surrounding buildings by a vision sensor in urban area

The existing GNSS (Global Navigation Satellite System) has the weakness in availability and continuity for the vehicles in urban area since the number of visible satellites decreases abruptly due to large buildings and other structures. In this paper, a new method is proposed to estimate the relative lateral distance of a vehicle from surrounding buildings by a vision sensor to supplement the shortage of satellite-based positioning. In the proposed method, it is assumed that a vehicle, equipped with a vision sensor and a low-cost inertial measurement unit, is moving along a straight road segment in urban area. The common feature points contained in the two successive image frames from the single vision sensor are extracted. Based on the common feature points, the lateral depth information from buildings is estimated and utilized to improve availability and continuity of position solutions. To verify the conceptual feasibility of the proposed method, a simple experiment was performed.